Security audit

Api Cost Optimizer

Security checks for vulnerabilities and agentic risk

Overview

This is a local cost-analysis skill that reads OpenClaw configuration and counts installed skill files, with no evidence of network exfiltration, persistence, or destructive behavior.

Install only if you are comfortable running local scripts that inspect your OpenClaw configuration and skill directory to estimate costs. Review reports before sharing them, and ensure required tools such as python3 and bc are installed; expect analyze.sh may need a small bug fix for the undefined tool-overhead variable.

SkillSpector

Vulnerability Patterns

MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 81% confidence
Finding: The documented purpose is narrow cost analysis, but the behavior reportedly includes scanning the local workspace and installed skill directories and reading multiple configuration locations, including per-agent files. That broader access increases the chance of unintended data exposure, overcollection, and operator surprise, especially because the extra behavior is not clearly disclosed; the broken code path also suggests insufficient review and raises reliability and safety concerns.

Static analysis

No suspicious patterns detected.